The White House is considering an executive order to establish an AI working group that would oversee pre-deployment testing of powerful AI systems. According to reporting from the New York Times and a statement from the Future of Life Institute, this working group would evaluate advanced AI models before they are released to the public, with a focus on national security risks.
This may sound like an American story, but it has direct implications for Irish businesses. The way the United States regulates AI will influence how the technology is developed, tested, and sold — and that affects every company using AI tools in Europe.
Why an AI Working Group?
The push for pre-deployment testing follows a series of wake-up calls. Anthropic’s latest AI model, codenamed Mythos, demonstrated capabilities that raised serious cybersecurity concerns. The company chose not to release it broadly, but the incident made clear that no single company should have the discretion to decide whether a potentially dangerous AI system reaches the market.
The proposed working group would sit within the White House and set standards for testing advanced AI models before release. This is similar in spirit to how new drugs must pass clinical trials before reaching patients, or how aircraft must be certified before carrying passengers.
What This Means for the EU AI Act
The European Union already has its own AI regulation framework — the EU AI Act. It classifies AI systems by risk level and imposes requirements on high-risk systems. But the EU framework focuses more on how AI is used than on how it is built. The US proposal would add a layer of pre-market testing that goes further than the EU AI Act currently requires.
For Irish businesses, this creates a situation where compliance requirements could differ significantly between the US and EU markets. If you use AI tools from US providers, those tools may need to pass US pre-deployment testing. If you use EU-developed tools, the requirements may be different. This regulatory divergence means businesses need to pay attention to where their AI tools come from and what standards they have been tested against.
The good news is that pre-deployment testing, if done well, should increase trust in AI tools. If you know a system has passed independent safety tests, you can use it with more confidence.
Practical Steps for Irish Businesses
The AI working group proposal is still at the discussion stage. It may take months or years to implement. But the direction of travel is clear: AI regulation is tightening on both sides of the Atlantic.
Here is what you can do now. First, audit the AI tools you currently use. Do you know where they were developed? What safety testing have they undergone? Second, keep records of how you use AI in your business. If regulation requires you to demonstrate compliance later, having documentation will save you time and stress. Third, talk to your AI providers about their testing practices. If they cannot tell you how their models are evaluated before release, that is a red flag.
The Bigger Picture
AI regulation is not just about preventing harm. It is about building trust. The proposed US working group and the EU AI Act both aim to create frameworks where businesses and consumers can use AI with confidence. For Irish businesses, staying informed about these developments is not optional — it is a competitive necessity. The companies that understand the regulatory environment will be better positioned to choose the right tools, avoid compliance pitfalls, and earn their customers’ trust. The message from Washington and Brussels is the same: AI oversight is coming. The businesses that prepare now will face less disruption when new rules take effect.